Model Test and Explanatory Power of the Model The ANOVA table for separate evaluation of the endogenous and exogenous variables as well as for the combined model shows the explanatory variable in the regression model are significant in explaining the drivers of profitability. In all cases, the calculated F value appears larger than the significance value. In other words the calculated significance value stood below 0.05. •ANOVA for internal variables- the calculated F-value(5.3) is higher than the significance value(0.04). •ANOVA for external variables- the calculated value (4.3) is larger than the significance value (0.17). •ANOVA for the combined model- the calculated value(2.9) is larger than the significance value(0.04). The residual statistics shows the error term has a normal distribution with a mean of 0. The results from Variation Inflation Factor (VIF) suggest that VIF is not greater than 10 for any of the explanatory variables. Hence, irrespective of the significance level of mulitcollinearity, it appears to be not serious and can be ignored. The variable with high VIF in the table is the logarithm of total asset which is statistically not significantly correlated with the independent variable. The Breusch-Pagan / Cook-Weisberg test for heteroskedasticity shows that at 5% level of significance, the p-value is higher showing that heteroskedasticy is not significant in the model. The Durbin Watson test result has shown that the D-statistic (2.8) appear closer but exceeds 2 depicting negative correlation. As suggested by Field (2009), values less than 1 or greater than 3 are a cause of concern1. Hence from Field’s rule of thumb it can be inferred that autocorrelation is not serious. 1The test statistic can vary between 0 and 4 with a value of 2 indicating that the residuals are uncorrelated. A value greater than 2 indicates a negative correlation and a value less than 2 depict a positive correlation.
European Journal of Business and Management ISSN 2222-1905 (Paper) ISSN 2222-2839 (Online) Vol.6, No.14, 2014 58 The coefficient of determination, the R and the R-square, is relatively high in all models. However, the combined (79%) and the endogenous (78%) models’ coefficient of determination by far exceed the coefficient of determination in the exogenous model (64%). Hence, the variable split in to endogenous and exogenous has indicated that bank specific variables are the better determinants of bank’s profitability in the Ethiopian context. External factors seem to have limited influence on the performance of commercial banks in Ethiopia. Both the endogenous (49.3%) and the combined model (41.9%) depicted a high explanatory power of each model. The reason for the reduced explanatory power in the exogenous variables is due to the dropped variables, in relation to serial correlation with the existing explanatory variables.